Detecting undiagnosed COPD: Using routine primary care data to select and validate candidate variables for COPD risk prediction

Abstract

Aim

To select and internally validate candidate variables to risk predict COPD in primary care.

Background

COPD is under-diagnosed in primary care. A validated algorithm could be used to identify patients who might benefit from spirometry.

Method

We used data on 17,719 COPD patients and 35,944 age/sex-matched controls, randomised in a 2:1 ratio to form derivation and validation samples. Candidate predictors were selected and adjusted ORs were estimated from a random intercept model. The AUC was estimated in the validation sample.

Results

Mean age in the derivation sample was 69.7 years and 51.8% were male. Smoking status, salbutamol use and dyspnoea were the strongest predictors. The model had an AUC of 0.875 (95% CI 0.87 to 0.88). A cutpoint of 0.3 yielded 86.6% sensitivity and 70.1% specificity.

Conclusions

Our interim model appears to be highly predictive of incident COPD and will be further developed and externally validated in a large screening trial (TargetCOPD).